Machine Learning Surrogate Modeling for Homogenization of Hyperelastic Materials with Boolean Microstructures
Researchers have developed a machine learning model to predict the effective properties of hyperelastic composite materials. This data-driven approach uses neural networks trained on microstructural descriptors, such as shape and correlation functions, to bypass complex numerical homogenization processes. The study found that while including more detailed descriptors like the lineal-path function improved accuracy at sampled points, it did not guarantee physically consistent behavior between those points, suggesting future work on physically constrained models. AI
IMPACT This research could accelerate material science simulations by providing faster surrogate models for predicting material properties.